More about the Programme
Although food-web modelling has been around in ecology for a long time already, it was only during the nineties that ecologists began to analyse the properties of food-webs in a consistent manner and developed tools that allowed the complexity to be expressed in a few summarizing parameters (Pim et al. 1991; Pahl-Wostl 1994; De Ruiter et al. 1995). This made the food-web approach an exciting new field of research in fundamental ecology. In general, one may distinguish between simple (strategic) and complex (tactical) food-web models (Scheffer 1990). Simple food-web models have a few variables. All trophic levels are typically considered to be in steady state except for one or two factors that are of interest. The number of parameters is low and default values may be provided by correlation with powerful characteristics of substances and species such as octanol-water partitioning and adult weight. Explicit solutions for the differential equations involved can usually be obtained. Since these models require relatively few data, they are successfully applied in cases where information is scarce (e.g. new chemicals), where the field situation is not very complex (e.g. accumulation of organochlorines in aquatic systems) and where predictions are urgently needed but only in terms of trends rather than exact concentrations or densities (e.g. phytoplankton control by zooplankton exposed to pesticides). Complex models have many variables and allow extensive non-equilibrium calculations, usually by means of numerical approximations of the equations (computer simulation). Their predictions are more accurate but at a cost of requiring many data. Complex models can give good predictions, but sometimes for the wrong reasons. In ecotoxicology, such models have been developed and calibrated for a few substances and ecosystems only.
The type of model needed depends on the question to be answered. In environmental management both simple and complex models are needed. They may also support each other: simple models can help to select cases to be studied in detail with complex models, results from complex models may be extrapolated to other conditions using more simple models, etc. The present proposal will therefore invite proponents of both type of modelling to participate in the programme.
The use of food-web models will allow the endpoints of ecological assessment to be defined better. With regard to the protection of ecosystems in the areas of environmental policy, as far as we know there have been no defined endpoints or target variables in terms of characteristics at ecosystem level. It is assumed that an ecosystem is sufficiently protected when all species inhabiting that ecosystem are protected. Rather than single-species endpoints, higher level targets should be defined such as:
• vitality (productivity)
• organization (ecosystem structure)
• resilience (powers of recuperation or toughness).
These variables conform closely to the general objectives of the European policy regarding the environment and nature conservation. In addition, target variables at the species level are relevant for the protection of designated species in a food-chain which are of particular value in terms of nature conservation policy. For these species endpoint could be defined as follows:
• average population size
• pseudo-extinction (likelihood of local extermination)
• resilience.
It is precisely these target variables that can be determined using the food-web approach. Classical food-web models are eminently well-suited to determine assessment endpoints at the ecosystem level. Hybrid food-web/population models can be used to determine such variables at the population level. Currently available food-web models cannot generate results in terms of species diversity and genetic diversity within species (policy objectives in the area of biodiversity). Further discussion is required to arrive at a selection of relevant target variables for the use of food-web models. In addition to the recommended general target variables, various specific endpoints should be considered. One such group consists of target variables relating to specific environmental functions, e.g. production, regulation and information functions. Another group consists of target variables arising from further elaboration of the biodiversity policy objective.
A programme should be developed comprising a series of international (primarily European oriented) round robin, and model shoot-out workshops (workshops in which various models are circulated among the different modellers, and tested by running a similar data-set) in order to combine already existing models which cover physico-chemical binding processes, physiological uptake and internal transfer processes, population biological, population dynamic, system structure and system process modelling. Characteristic in this context is the multidisciplinary co-operation of physico-chemists, physiologists and biochemists, and ecologists (Peijnenburg et al. 1997).
Although a number of well known and well studied toxic compounds seem to be reasonably controlled by environmental policy, the following aspects are still insufficiently investigated with respect to persistent and toxic compounds like heavy metals, organochlorines, and PAHs:
1. The total mixture of combined impacts of low level persistent compounds are hardly understood, let it be properly assessed;
2. Quantitative assessment of the ultimate fate of a persistent compounds when transferred in an ecosystem is only partly (for some compounds) developed;
3. In this assessment, the real fate (including dynamic sorption in the environment, internal degradation and sorption in target organisms and their organs, and biodegradation by micro-organisms) has been included only incidentally;
4. Proper ecological assessment of the potential impact of new compounds is still weakly developed (this could comprise compounds with a very specific mode of action like endocrine disrupters or compounds with a very specific transfer route like cadmium uptake via willow bark to beavers;
5. Indirect impacts by interfering food-chain or more complex relations (chemical signals) are still hardly studied;
6. Food-chain models which describe these kind of phenomena are promising tools to describe long term implications of these phenomena;
7. These have to be combined with a. an ecological analysis in the field of the real transfer routes and target species and b. a literature study about the potential physicochemical availability under variable (dynamic) conditions and resulting (eco)toxicological processes.
Of course this list of problem areas is far too complex to cover in one programme. By choosing a few representative compounds from these groups we can cover nevertheless a number of these problem areas. Moreover, ecosystem models provide an effective tool to assess the combined impact of the combined set of contaminants.
In a recent report the Standing Committee on Ecotoxicology of the Health Council of the Netherlands has reviewed the state-of-the-art of food-web modelling (Health Council 1997). From this review on the international status of food-web modelling the following concluding observations can be derived.
Food-web models are effective aids to a better understanding of the consequences of disturbances to these relatively complex relationships. The classical food-web models enable predictions to be made concerning the effects of toxic substances on ecosystem processes such as nutrient cycling and energy flows. This is because feeding relationships play an essential part in such processes. With such models, predictions can be made concerning effects on structural features of ecosystems such as their stability and resiliency. With hybrid food-web/population models, which combine classical food-web models with detailed population models, predictions can be made concerning the effects on the populations of certain individual species (in terms of population densities and likelihood of extinction).
Relative to work carried out in the area of aquatic ecosystems, food-web research and model-building in relation to terrestrial ecosystems are lagging behind. Currently available terrestrial food-web models are restricted to a few agricultural systems. They usually deal only with interactions in the subsurface (detritus web). The Health Council committee recommends that such models be expanded to include interactions with above-ground elements of the terrestrial food-web (plants, herbivores and the like). Also in the aquatic-sediment models this detritus cycle is of great importance to model the fate and impact of contaminants.
In the Committee’s view, an important obstacle to the use of food-web models in ecotoxicological risk assessment is the lack of hard data, or more precisely a database consisting these data. This includes ecophysiological and life-history parameters (data on consumption, mortality and the like, as well as data on concentration-effect relationships for those parameters) for the various functional groups in an ecosystem. It is not sufficient simply to apply extrapolation methods to a limited amount of toxicity data. In various ecosystems rather specific environmental conditions occur which have a major impact on the fate and behaviour of different contaminants. Moreover, it is important that the same set of default values and basic biological parameters is used in order to make the output of the various models properly comparable. This is the more important when models of aquatic and terrestrial ecosystems are coupled to assess consequences of pollution in the framework of environmental outlooks.
The provision of data for food-web models is often a bottleneck. We believe that it is not necessary per se to collect a vast amount of new data before food-web models for risk assessment can be developed. The data needed as input for food-web models can be found scattered in the literature. For example, in soil invertebrate research there is a wealth of (partly internal) research reports providing the data as mentioned above (Posthuma, pers comm, see also Sheppard et al. 1998; Axelsen et al. 1997). What is needed is the creation of a database in which these data can be brought together in a consistent way.
Another significant bottleneck is that the results of most ecotoxicological food-web models have not been validated against field data. A problem here is a lack of information on the densities and biomasses of major groups of species in contaminated and uncontaminated ecosystems. This validation against field data, which deserves high priority, is not the only way to obtain insight into the uncertainties surrounding the results obtained from models. Tools such as sensitivity analysis and uncertainty analysis can be used to resolve this issue. Other approaches include validation of individual components of food-web models, for example the parts which describe major sub-processes. Pending the results of validation tests, sensitivity analyses and uncertainty analyses, no well-founded judgements can be put forward concerning the reliability of results obtained using currently available food-web models.
Again, the major limitation in this context is the existence of a proper data base bringing together the detailed information about a number of extensive field studies and inventories carried out in the past decade; examples are the Swedish Rönnskär and Gusum areas, the English Avonmouth area and the Dutch-Flemish Kempen area, all contaminated with considerable amounts of heavy metals. An attempt will also be made to validate model output using data from the Doñana case and other environmental disasters in Europe.
Supplementary research in order to use food-web models for ecotoxicological risk assessment involves the further development of food-web models (modifying and extending their structure), especially models describing natural, complete terrestrial ecosystems. It also involves collecting the required input data and parameter values either in the laboratory or in the field. Furthermore, a comprehensive set of field data is required for validation purposes. With respect to this last conclusion in The Netherlands and Germany, programmes have been initiated which aim at a multidisciplinary monitoring programme on contaminant inputs and effects in different ecosystems. In the Netherlands the first series of projects within this programme on "Ecosystem Oriented Ecotoxicological Research" will start effectively January 2000. This programme has ecosystem-modelling as a core and leading activity.
Summary of goals and expected achievements
The programme aims to stimulate the development of food-web models for risk assessment of pollution in the terrestrial environment. It could provide a "trait-d’union" between monitoring and validation-oriented ecotoxicological programmes currently running in European countries. A further generalisation of the data derived by these studies to other temporal and spatial scales by application of system modelling is expected. There will be a significant spin-off to fundamental studies in community ecology. Expected concrete outcomes of the programme may be listed as follows:
• An ecological data base supporting food-web modelling in the terrestrial environment.
• An ecotoxicological data base involving toxicity data for terrestrial species.
• A selected number of food-web models for representative European ecosystems.
• A system of higher-level ecological evaluation of potentially toxic chemicals.
• A manual for food-web modelling in the terrrestrial environment as related to environmental pollution, as well as several other books and scientific articles.
Activities
Workshops/Summer Schools - Call for proposals [CLOSED]
Conferences
Grants - Call for applications [CLOSED]
Databases
Email mailing list
Links
A brochure containing information on the scientific background and aims and objectives of the programme has been published. The brochure can be downloaded by clicking here (PDF 304 KB)
Links
CLARINET - The Contaminated Land Rehabilitation Network for Environmental Technologies in Europe Go to Website
